On Fri, Aug 27, 2010 at 04:03:20PM +1000, matthew green wrote: > not sure exactly what that means here... but maybe acpicpu should > default to blacklisting P4's? for when it becomes a normal driver > that is enabled by default..
Blacklisting based on processor type sounds reasonable, but somewhat problematic. We can safely let acpicpu(4) attach but disable specific functionality like P-states. However, how sure can we be that ICH-based SpeedStep was available only in P4s? I also think that at least some P4-models (Prescott?) want to use ACPI. Thus, I think the real issue is more about piixpcib(4) rather than ichlpcib(4). > FWIW, acpi_cpu doesn't work for my P4 with speedstep, and all sorts of > things are missing without acpi (including stability under load.) it > has no old-skool APM (unlike most laptops from that era.) Pentium 4 is not really that old; ACPI is certainly present. If you send (off-list) the acpidump(8) output to me, we can ensure that this is not a firmware bug specific to this laptop, or a bug in acpicpu(4) itself. - Jukka.
